xlsr_ur_training
This model is a fine-tuned version of facebook/wav2vec2-large-xlsr-53 on the common_voice_8_0 dataset. It achieves the following results on the evaluation set:
- Loss: 0.8325
- Wer: 0.4863
Model description
More information needed
Intended uses & limitations
More information needed
Training and evaluation data
More information needed
Training procedure
Training hyperparameters
The following hyperparameters were used during training:
- learning_rate: 0.0001
- train_batch_size: 8
- eval_batch_size: 8
- seed: 42
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- lr_scheduler_warmup_steps: 1000
- num_epochs: 30
- mixed_precision_training: Native AMP
Training results
| Training Loss | Epoch | Step | Validation Loss | Wer |
|---|---|---|---|---|
| 6.9537 | 3.25 | 1000 | 3.0940 | 0.9989 |
| 2.1696 | 6.49 | 2000 | 0.9705 | 0.6830 |
| 0.8637 | 9.74 | 3000 | 0.8098 | 0.5919 |
| 0.6297 | 12.99 | 4000 | 0.8002 | 0.5469 |
| 0.5034 | 16.23 | 5000 | 0.8019 | 0.5214 |
| 0.4267 | 19.48 | 6000 | 0.8223 | 0.5085 |
| 0.3847 | 22.73 | 7000 | 0.8081 | 0.4948 |
| 0.342 | 25.97 | 8000 | 0.8300 | 0.4930 |
| 0.3201 | 29.22 | 9000 | 0.8325 | 0.4863 |
Framework versions
- Transformers 4.21.0
- Pytorch 1.11.0+cu113
- Datasets 2.4.0
- Tokenizers 0.12.1